ComfyUI Local Install and ComfyUI Manager On Apple Silicon M1/M2/M3 Mac Full Tutorial
Summary
TLDRビデオでは、MacにComfy UIをインストールし、その操作方法を紹介しています。Comfy UIは、より高度なAI画像生成ソフトウェアで、多くのオプションを提供し、ビデオ生成にも対応しています。インストールにはHomebrewとPythonが必要ですが、手順は比較的簡単です。Comfy UIを起動し、モデルを読み込む方法や、ポジティブ・ネガティブプロンプトを接続して画像生成を行うプロセスが解説されています。さらに、Comfy UIマネージャーを使用して、モデルのインストールや更新、ワークフローの再現も可能です。
Takeaways
- 😀 コミュイ(Comfy UI)はMacにインストールすることで、より高度なAI画像生成ソフトウェアとして自動1111を超越することができる。
- 🛠️ MacにComfy UIをインストールするには、HomebrewとPythonが必要なセットアッププロセスを経て、ターミナルを通じてコマンドを実行する必要がある。
- 🔗 安定性拡散(Stable Diffusion)を使用する場合、pytorchというディープラーニングフレームワークが必要なため、Macにインストールする必要がある。
- 📚 Comfy UIは複数のオプションと機能を提供しており、画像生成において高い柔軟性と制御が可能である。
- 🎥 Comfy UIはビデオ拡散にも対応しており、今後ビデオに関する説明動画が提供される予定である。
- 💻 Comfy UIはMac上で直接実行でき、Windowsと比較して異なるインストールプロセスを経る。
- 🔄 Comfy UIマネージャーを通じて、モデルのインストールや更新、欠落しているカスタムノードのインストールが可能である。
- 🖼️ Comfy UIを使用することで、特定の画像を再現するためのワークフローを視覚化し、再現性を高めることができる。
- 🔍 Comfy UIは高度な設定オプションを提供しており、ユーザーはそれらを調整して生成される画像の品質やスタイルを制御することができる。
- 🌐 Comfy UIに関するリソースとして、GitHub、Reddit、チュートリアルウェブサイトなどのコミュニティとサポートが存在する。
Q & A
なぜMacでComfyUIをインストールする必要があるのですか?
-ComfyUIは、Mac用の画像生成ソフトウェアとして非常に優れており、Automatic 1111よりも多くの制御オプションを提供します。
ComfyUIのインストールを始める前に必要なものは何ですか?
-HomebrewとPythonが必要です。
Homebrewをインストールする方法を教えてください。
-Homebrewの公式サイト(brew.sh)にアクセスし、提供されるコマンドをターミナルに貼り付けて実行します。
Pythonをインストールする際に推奨されるバージョンは何ですか?
-Python 3.10または3.11が推奨されます。
PyTorchのインストール手順を教えてください。
-PyTorch公式サイト(pytorch.org)にアクセスし、プレビュー・ナイトリー版を選択してインストールコマンドをターミナルで実行します。
ComfyUIのインストール方法を教えてください。
-ターミナルでgit cloneコマンドを使用してComfyUIをクローンし、ディレクトリ内でpip3 install -r requirements.txtコマンドを実行します。
モデルをComfyUIに読み込む方法は?
-モデルファイルをComfyUIディレクトリ内のチェックポイントフォルダに移動し、ComfyUIを再起動します。
ComfyUIで画像生成を開始するにはどうすればよいですか?
-ターミナルでpython3 main.pyコマンドを実行し、表示されるHTTPアドレスをブラウザに入力してComfyUIにアクセスします。
ComfyUIのノードとは何ですか?
-ノードは、チェックポイント、プロンプト、サンプラーなどの画像生成に必要な各種設定を表す要素です。
ComfyUIマネージャーの役割は何ですか?
-ComfyUIマネージャーは、モデルやノードのインストール・管理を容易にし、プログラムの機能を拡張します。
ComfyUIを他のシステムと共有する方法は?
-ComfyUIで作成した画像をドラッグ&ドロップしてワークフローを表示し、他のユーザーが同じ設定で画像を生成できるようにします。
ComfyUIでモデルパスを変更する方法は?
-テキストエディタでextra_model_paths.yaml.exampleファイルを開き、適切なパスに変更して保存します。
ComfyUIのアップデート方法を教えてください。
-ComfyUIマネージャーで「Update ComfyUI」または「Update All」を選択してアップデートします。
ComfyUIでノードをインストールする方法は?
-ComfyUIマネージャーの「Install Custom Nodes」を使用して、必要なノードをインストールします。
ComfyUIの設定を変更するにはどうすればよいですか?
-ComfyUIの設定メニューで、使用するカラースキームやその他の設定を変更できます。
Outlines
😊 ComfyUI for Mac Installation Overview
In this introductory paragraph, Chuku Bum explains the advantages of using ComfyUI over other AI image generation software on Mac, such as Automatic 1111 and Mid Journey. He highlights ComfyUI's robustness, control over image generation, and simplicity despite initial intimidation.
🔧 Installing Homebrew and Python
This paragraph guides users through installing Homebrew on their Mac, explaining the steps to download and verify the installation using the terminal. It also covers installing Python, recommending additional tools like cmake and git to support future use of Automatic 1111 alongside ComfyUI.
💻 Installing PyTorch and Cloning ComfyUI
Chuku Bum explains how to install PyTorch, a crucial framework for deep learning models, necessary for running ComfyUI. He then walks through cloning the ComfyUI repository into a preferred directory on the Mac, making it accessible and ready for further configuration.
📂 Configuring and Running ComfyUI
The paragraph details the process of setting up ComfyUI by installing required dependencies and running the main script to launch the interface. Chuku Bum emphasizes the need for a model checkpoint to generate images and demonstrates how to download and place the stable diffusion model in the appropriate directory.
📈 Understanding ComfyUI's Interface
Chuku Bum provides a crash course on using ComfyUI, explaining the roles of different nodes (e.g., checkpoint, prompt, sampler) and how to connect them. He discusses using the ComfyUI Manager for downloading models and custom nodes, showcasing the interface's flexibility and control.
🔄 Advanced Features and Resources
The final paragraph highlights advanced features like importing workflows, updating ComfyUI, and integrating models from Automatic 1111. Chuku Bum suggests resources for further learning, including GitHub, Reddit, and specific websites, encouraging users to explore and optimize their AI image generation setup.
Mindmap
Keywords
💡ComfyUI
💡Homebrew
💡Python
💡PyTorch
💡Terminal
💡Git
💡Models
💡Nodes
💡Custom Nodes
💡Sampler
Highlights
Chuku bum introduces Comfy UI as a superior AI image generation software for Mac, offering more control and options compared to Automatic 1111.
Comfy UI enables stable video diffusion and various other features, setting it apart from competitors like mid Journey.
The presenter emphasizes Comfy UI's ease of use and long-term learning benefits over Automatic 1111.
A step-by-step guide begins with the installation of Homebrew and Python, prerequisites for Comfy UI.
Instructions are provided for downloading Homebrew from brew.sh and installing it via the terminal.
The tutorial recommends installing additional tools like cmake, protuff, rust, and git for broader capabilities.
PyTorch, a framework for deep learning models, is highlighted as essential for Comfy UI's functionality.
Detailed steps are given for installing PyTorch from pytorch.org using the appropriate command in the terminal.
Comfy UI can be easily installed on Mac by cloning the repository, differing from the Windows installation process.
The presenter shares a personal challenge with the installation process and credits AI animation for assistance.
Comfy UI's interface is explored, explaining the nodes and their functions in the image generation workflow.
A demonstration of generating an image using Comfy UI is provided, showcasing the software's capabilities.
The importance of selecting the right model and its impact on VRAM usage is discussed.
The process of adding a model to Comfy UI and using it to generate an image is explained.
Comfy UI Manager is introduced as a tool for organizing and enhancing the software's capabilities.
The ability to install models and custom nodes directly through Comfy UI Manager is demonstrated.
Comfy UI's potential for real-time image generation with the addition of certain models like sdxl turbo is highlighted.
The presenter guides on how to use Comfy UI Manager to import and replicate previous workflows.
An overview of Comfy UI's settings and customization options is provided, including color schemes.
Resources for further learning and community engagement with Comfy UI are shared.
The video concludes with a summary of Comfy UI's benefits and a call to action for likes and subscriptions.
Transcripts
what is going on ladies and gentlemen
it's your boy chuku bum and I am here to
talk to you about how we're going to get
comfy UI installed on Mac and why would
you need comfy UI installed on Mac well
first off it's a much better AI image
generation software than automatic 1111
for Mac it's a very robust and probably
the best image generation software for
Mac as of right now you have way more
control over the types of images that
get created and allows for even more
options when it comes to image
generation let alone stable video
diffusion which you can expect a video
on very soon plus so many other options
for you that make it way better than
just using automatic 1111 or you know
maybe mid Journey or whatever the case
and it'll be right here on your Mac it's
just a win-win and although it can seem
a little daunting at first it's actually
very simple to use and in the long term
way better to get to learn than even
automatic 1111 would be so let's just
hop right into this and get started the
setup is going to be very similar to
what we did with automatic 1111 for Mac
with stable diffusion you're going to
need Homebrew and you're going to need
python so first we need to download home
brew so we're going to go to brew. sh
boom now you're at home brew and then
what's going to happen is you're going
to take this right here you could just
click this little copy button right here
you're going to take this copy button
and you're going to head to your
terminal if you don't know how to head
to your terminal you can click f4 on
your Mac and then you can type terminal
and then you'll get over there very
quickly and all you're going to do
you're just going to command V your this
bad boy you're going to paste it right
into here now I already have Homebrew so
it's pretty much just going to stay
installed but um you're just going to
run it and then you're going to say
checking for pseudo access D so you're
going to put in your laptop's password
it says this script will
install and then you press enter that's
perfectly fine now like I said I already
have it so it's going to just update
everything pretty much installation was
successful and all it did was update two
tabs but for you it would be a whole
thing like it would be like and then if
you want to make sure that you have home
brew you would just type Brew doctor
after you type in Brew doctor and then
it'll let you know if you have home brew
so he says your system is ready to brew
so that's cool that even they have a
nice sense of humor because you know
Brewing brewing a beer a brew you get it
now that we've installed Brew it's time
to install python now you could install
python alone that would be perfectly
fine and you would be able to get comfy
UI up and running with the next steps
that we provide but I want to give you
guys a little bit extra because you may
also in the long run want automatic 1111
so what I'm actually going to recommend
to you guys is that you come over to
this website right here the link will be
in the description below and what you're
going to do you're going to do Brew
install cmake protuff rust and python U
see this python is 3.10 you could always
do 3.11 that's the newest python that's
the newest stable python version and
then you know um Brew install git and
Brew install w G now these are all
things that actually will give you the
ability to run automatic 1111 rather
than just installing Python and then all
you can do is comfy UI so I'm definitely
trying to give you a little bit more
than what you would normally get but if
this is a little bit Overkill you could
just get rid of everything here and just
do Brew install python at 3.10 or 3.11
or you could just leave the threes alone
and just Brew install Python and you
would get Python and you could still
continue on with this tutorial but I
definitely do recommend that you take
this whole thing right here so I'm going
to take this whole thing and just to
show you what I mean when I say that you
can always just get rid of the rest
right if you want you could literally
just do Brew install python if you run
this that'll be perfectly fine in fact
I'll do it really quickly if we do Brew
install python see I already have python
so it's going to say python 3.11 more
specifically 3.1.6 one is already
installed and up todate so I don't need
to install anything it's asking me if I
would want to reinstall I don't need to
do that but you could run that and you'd
be good to go but me I'm going to run
the whole thing just to show you guys
again Brew install cmake Proto and
everything else because again just
python will work for comfy UI but if you
install everything else you'll also be
able to use automatic 1111 if you ever
decide to have that so I always like to
say better to have and not need than to
need and not have so we're going to run
this again Brew install C make Proto
above everything and again I already
have everything so it's just going to be
you know reinstall reinstall reinstall
but you guys would get like this long
list of stuff that's what you guys would
get and then you'd be good to go so now
that you've installed python along with
other things now it's time to install
pytorch and pytorch well look at it
right here pytorch is a fully featured
framework for building deep learning
models basically meaning that this is
what helps create chat gbt stable
diffusion mid Journey claw Bing AI all
these different AIS that we're getting
more and more used to seeing on a daily
basis it's built a lot by using py P
torch and so we need pytorch on our Mac
in order to get comfy UI up and running
now in order to get pytorch you're going
to come here to py torch. org get
started SL locally you're going to come
over here and you're going to click
preview nightly that's the type of pie
torch that we need that's going to run
on Mac so now we have preview nightly
again the OS is Mac the package is PIP
um the language will be Python and the
compute platform will be default I mean
it's pretty much already selected for
you the only thing that you got to
change is nightly and then once you have
this you're going to make sure that
you're going to uh take this right
you're going to take this code right
here copy it however you'd like and
you're going to come into this terminal
after you've run everything and you're
going to put pyour just like that and
it's going to run okay now again I have
it already I don't know I'm G that's
there going to be a theme in this video
I already have that so it says already
already installed already satisfied um
you guys would get something different
you get you guys would get like a whole
bunch of loading things and everything
like that but with me I'm getting
already satisfied so now we have Brew
Python and pytorch so we're just about
ready to get the ball rolling now that
we have everything that we need we're
going to install com fui onto our
MacBook and unlike trying to install com
fui on a Windows where you would have to
get clone comfy UI onto your windows and
then use a seven zip extractor to get it
out and do all this stuff with U Mac
it's actually really simple what you're
going to do you're just going to copy
this git clone right here and you're
going to pretty much CD into whichever
directory that you want to be in so
right now by default in my terminal I'm
in the root of my Mac so you can see you
could tell that by this little sign
right here this little like curvy sign I
forgot what that's called but if you
want to know where you are in your Mac
all you do is type
PWD and it'll show you I'm in the users
chukua directory right but what's in
that directory all you would do is type
LS and it will list everything that's in
that directory now me personally I'm
going to clone comfi onto my desktop you
can do do it into your downloads you can
do it into your documents folder you can
do whatever you'd like but I personally
love to put things on my desktop because
it's just easy access quick access so
I'm going to now move into that
directory by typing in CD and then
desktop boom now I'm in desktop and
again you can always check by doing PWD
and it says it right there users chub
Maga desktop so now we're going to get
clone comei onto our desktop and all you
have to do is command V now if you want
to make it more sturdy people would
probably get clone comi into another
folder because sometimes you may want
like all your AI stuff in one folder
like let's say automatic 11 comfy UI all
within an AI image generation folder but
for now I want to keep it simple and
just come and just clone it right there
onto my desktop so now that we have it
in our terminal we're in our desktop
we're going to say get clone we're going
to press enter and it's going to start
cloning in and it's pretty quick and you
could tell that it's there now because
you'll see comfy UI right here now that
we have comfy UI installed we're going
to need to run one more command but
before we do that I do want to show you
guys what the comfy UI would look like
so again if you open the comfy UI folder
you're going to see all these things and
this might say look a little daunting at
first but they all mean something and
it's not that hard I I'll teach you guys
and we'll walk through it but before we
go through all this there is one command
that needs to be run and I'm not going
to lie I actually had a hard time once I
got to this part of the installation
process and I would have to shout out AI
animation they really helped me with a
lot of this stuff right here AI
animation great YouTube channel talks
about all types of AI stuff and um they
actually literally they they actually
pretty much solved the issue here there
weren't a lot of there wasn't a lot of
info about how to get past this part or
at least get past this part correctly
and these guys really help me figure
that out so this this video probably
wouldn't even be possible as soon as
it's being made if not for AI animation
so definitely go show them some love I
got to really appreciate them but now
because of them we have the next step
and we're going to get even deeper like
I said now we have com fui here on our
desktop so now we need a CD into desktop
so we're going to CD into
desktop great we've cded into desktop
and what's funny is that as we're moving
through each folder path you'll see this
top part change it'll change into
wherever you are so that's a good way of
knowing where you are along with the PWD
command so now we're going to uh CD into
comfy UI you don't have to worry about
capitalization or anything it'll
automatically just go straight into that
one so CD into comi but again if you
never know it says it right here comi
and you see how I didn't have to worry
about capitalization uh Capital C
Capital UI it's it's the same exact
thing there and again if you ever want
to know where you are PWD and you'll
find it right there and you know one
more tip and trick if you type in CD
Dash it'll take you back into the
previous directory so now I'm back in
the desktop directory and you can see
that at at the top right here uh
reflecting that chain but we're going to
go back into com fui and we have to run
this command pip 3 install dasr
requirements. text now what does this
command do well if you go back into this
comfy UI folder you'll actually see it
right here requirements. text and if you
open that up it'll show you what is
actually needed to run comfy UI and
basically all that is going to do is
that once you run that command it's
going to run that script to get what's
needed for comfy UI so again we're just
going to ex out out of that and we're
going to run the command simple as that
so you see now again like I said I've
run this all before so you're going to
I'm getting requirement already
satisfied um but because it's already on
my Mac but some but you guys would get
like long loading things and everything
like that but it's still not that long
usually takes about a minute or two so
now we have comfy UI on our laptop Mac
whatever and let's run it while still
being in this directory you're going to
type in Python on 3
main.py okay now that that's just
activating this file right here that
actually runs comfyi so once you press
enter on this command it's going to
activate comfyi it's going to take a
little bit of time but it's going to
activate comfi and once comfyi is
activated you're going to take this HTTP
address right here you're going to copy
that bad boy you're going to paste that
bad boy right here and boom you have
comfyi ready to go oh actually not
really you don't have comfy UI ready to
go because you actually don't have any
models to use so look if you click right
here you need to load a checkpoint a
Model A Safe tensor you need this in
order for it to work and as you can see
right now it says undefined so if I try
to queue a prompt it will not work it'll
actually give me prompt outputs failed
validation so we need to go get a model
now when it comes to models different
models will actually require different
levels of vram so I'm actually going to
get like a nice chill model so that we
can run this very fast and very
effectively so the model we're going to
download for this one is a very chill
very usable stable diffusion v15 model
um obviously there are much better
models but it's it's just a nice stable
model that typically does not give a
hard time and it's still very fast to
use across all systems and we're going
to download the v15 pruned emo only safe
tensors be very wary about downloading
models with checkpoints checkpoints are
pretty much safe tensors they're pretty
much the same thing it's just that
checkpoints can carry malicious viruses
it can C it can carry malware and you
really do not want that so you want to
take the safe tensor route normally and
yeah you would just take that and you're
going to download it it'll take some
time a couple of minutes and once it's
downloaded I'll show you what to do next
awesome so now that it's downloaded
what's going to happen is that we're
going to open two finder windows or you
can open one I mean you can kind of do
however you want but here we have the
down downloads folder right you you're
going to navigate to where the downloads
folder is on one of them and you can see
where we have it v15 pruned Emil only
save tensors right now I need you to
navigate to your comfy UI folder and
then in that folder you're going to get
into the models folder and in that model
folder you're going to get into the
checkpoints folder and right here you're
going to say put checkpoints here now
again just to clear up any confusion a
checkpoint and a model and a save tensor
are all the same they're just templates
that stable diffusion uses to base the
images that you're generating off of it
can get much deeper than that but that's
basically what it is and you're just
going to move
this this into that there you have it so
now we have it in our checkpoints okay
great now you're going to come back to
where comfy UI is now you're going to
need to refresh this so you could this
is literally a web page so you could
literally um you know command R and
refresh it or you could click refresh
here perfectly fine either way that you
want to do it and now you'll see that
v15 uh prune EMA only save tensors is in
there so now we finally have a model and
we have our prompt beautiful scenery
nature glass bottle landscape purple
Galaxy bottle and then we have our
negative prompt which is text Watermark
now just to show you a little bit of how
this works well actually before I get
into how this works let's actually run
the prompt now that we've run the prompt
a few things that you can know is that
if you you can always just grab the
screen and move it around um on your
MacBook using your trackpad you just you
know scroll up and down to zoom up and
down comfyi typically much faster than
if you were to use automatic 1111 but
still know that Mac in general when it
comes to stable diffusion is a lot
slower than a good PC so there's still a
ways to go when it comes to stable
diffusion on Mac but using com fyuui
will on average be a lot faster than
trying automatic 11-11 and look wow look
at that look at what we got you know
look we got this nice little thing and
boom now you've generated your first
ever image so this is really great um
you don't have to just scroll in like
that you can also click on this and then
extend the image like so now we have our
first ever image but there's a few
things that you need to know about comfi
so this is going to turn into a little
bit of a crash course so first and
foremost this is where you load in your
checkpoint and your model from your
checkpoint and your model you have to
connect it to a prompt a positive prompt
and a negative prompt and you connect it
using these noodles so all you have to
do is take this little dot you would
click it and then you would connect it
to here or connect it to there and stuff
like that and that's basically what
happens with the clip the clip connects
to the positive and the negative prompt
and then the positive and the negative
prompt they connect well well
technically speaking neither of these
prompts are actually positive or
negative they're not determined until
these prompts are connected to the
positive and negative sources in this
sampler so you take this and it would
connect to negative and you would take
that and connect to positive so you can
actually literally see this I can
actually disconnect it so you see I just
disconnected it so now again they're
just prompts you don't they don't mean
anything until you finally connect it to
a positive or negative and you can
connect the bottom to positive or the
bottom to negative it doesn't really
matter just obviously for organizational
purposes negative is on the bottom
positive is on top then you have the
model now the model ption portion of
this connects to the sampler itself and
the sampler is what runs your image
generation the sampler is what kind of
creates the code and the process that
creates the image that we're seeing over
here right that's what the sampler does
and it has a lot of different settings
to it that oh we need not go into right
now but the sampler has a lot of
different settings and the more that you
adjust these settings the more that you
get and then you also have this latent
image and all this does is determine the
size of the image so if I change this
from you know uh 10 to 1080 you know
1080 by 1080 you would get a much bigger
picture of course but we're not going to
do that either right now and then this
connect to the decode and the decode is
what takes all the information and make
and packages it nicely to create the
image and place it right here so that's
kind of what this whole process is and
it's a set of noodles and all these are
called nodes by the way so they're
different nodes this is a checkpoint
node a prompt node a sampler node but
they're all nodes just for you to know
and com fui can get way more Swifty but
in order for it to get more shwifty
you're going to need one more thing
which is the comfy UI manager that helps
organize this whole stuff and add much
more to this and make this a much more
robust program than what we're seeing
right now so let's get into that so
you're going to come over to this link
and again everything in the video will
be linked down below the way that we're
going to do it there's a couple ways to
do it um but I like this way the most
cuz again when you learn the terminal
it's even better um what happen is that
you're going to CD into custom nodes so
we're going to actually show you another
nice way to do this see custom nodes let
me show you where custom nodes is custom
nodes is just another folder in your
comfy UI directory so you got to get
into custom nodes right now the thing is
that in order to get into custom nodes
yes you could do the CD into CD into CD
to get there but what you could do also
is that you can go to this terminal
right now what we're going to do
actually we're going to open up a new
terminal window just open up a new
window and what's going to happen now
you could now again you could actually
close the com UI if you want but I'm
just going to leave it running and we're
going to open up a new window and what's
going to happen is that we're going to
type CD so we're going to change
directory that's what CD stands for
change directory and you're going to
drag custom nodes in here and it's
actually going to copy the path file so
um or rather the path name and then you
just click enter and now you have cded
into custom nodes and again you can see
it right here custom nodes is right here
and you're going is going to go back to
the link and you're going to take this
whole thing get clone you're going to
get this whole thing get clone HTTP
GitHub and then now that you're in here
you're going to get clone it into custom
nose that
folder and you see it's going to clone
comfy UI into it so remember we still
have comfy UI running so now that we've
cloned The Comfy UI manager into comfy
UI you are going to need to stop the
comfy UI software from running and then
restart it so in order to do that all
you got to do is just X Out the terminal
that is running in pretty simple you
just exit out and that will terminate
comfy UI like as you can see right here
you'll see that it now turns into into
reconnecting because the terminal is no
is no longer running so we just got to
make our way back into the comfyi folder
so you again you can use the same method
CD into comy UI like I just did I
dragged and dropped it over there and
then we're going to do Python 3
main.py again and it's going to start
running everything up just like normal
and you're going to take that link and
you can you can actually highlight this
then double click it and then um open
link and it'll just open default into
the same same thing and then boom now
you're going to see something different
you're going to see manager and now
we're cooking with oil cuz check this
out now that you have the comf UI
manager there's going to be a lot of
things that you can do so first off you
can install models so you see how we
were taking models before we had to go
to hugging face or civid AI in order to
get models you don't need to do that
anymore at least for some models and
some nodes and some ups scalers and
things of that nature you can literally
just download them into comfy UI and
them from there so for example right
let's close this for a second as you can
see let's close this as you can see we
only have one model to use v15 pruned
okay great when you go into
manager and you click install models
right there are different types here
there's upscale models that's for making
the the the face better there's face
restore models that help to rebuild the
face if the AI gets it wrong and so on
and so forth and then there are
checkpoint points this one is for stable
diffusion video models which we will get
into um on another video but then there
are some St some regular ones like
checkpoints that you can use like sdxl
turbo there's also control net there's a
lot of things that you can do here but
for now we're actually going to keep it
very very simple we're going to download
just another version of a model so check
this out all you would have to do let's
see sdxl turbo we would install that and
if you want to learn how to use sdxl
turbo which allows for realtime image
generation you can actually go check out
this video here where I talk about it
and set it up with you so it's pretty
big stuff and by the way if you're
liking the video so far and I've been
able to help you please remember to like
And subscribe it takes this channel to a
whole another level and I'd really
appreciate it thank you so now you can
see that it's installing sdxl turbo one
it'll take some time but once it's
installed it'll just say installed and
then we can move forward and you'll see
it right there in the model checkpoint
and boom now it's installed to apply the
installed model please click the refresh
button on the main menu so we're just
going to click close then we're going to
again and close this we're going to
click refresh now that it's been
refreshed if you click on this you see
sdxl turbo and we could change it but
we're not going to use that right now
we're going to stick with the uh v15
safe tensors so that's how you're now
able to install models into your system
another cool thing that you could do
when it comes to comfy UI you can bring
images that were created in comfy UI
into your own comf UI and it will
literally show the workflow that made
that so I'm going to show you right now
with one of the images that I made a
while back ago so here I have an
attractive black woman that I had made
and so if I drag and drop this into the
UI you see that it just changed the
whole workflow you can literally see
what I did to create the black woman
you'll see that it used what I created
to make the prompt you'll see that uh it
had dream Shapers which I don't have
right now this is a different comy UI um
you see that I also had used an upscaler
and up I had used a couple ups scalers
and stuff now if I cue The Prompt it
won't work because look see prompt
outputs validation you know the
checkpoint loader it doesn't have that
checkpoint and so on and so forth so as
a result it can't use it I'll need to
download these things now the beautiful
thing is that it shows you what you need
to download so I would just need to get
the dream shaper safe tensors and says
value not en list model name forx anime.
paath so I would need to get this as
well and then you know once you're able
to get those you would be able to move
forward so that's the beauty of it you
can bring an image that was made in
comfi into your own comfyi so that you
can see the workflow that was used to
create that same exact image and try to
replicate it yourself which is very
powerful when you're trying to have as
much control as possible over the images
that you're trying to create another
thing about the comfy UI manager is that
whenever you want to update comfy UI
itself you can literally just say update
com UI or update all meaning that you
would update the nodes the models
everything which can also help you out a
lot you can also install missing custom
nodes basically meaning that whenever
you do take someone's workflow and it
has all types of nodes um that you don't
have you can install it doing that way
and if you already have automatic 1111
here's a nice cool tip that you could
use instead of R downloading every model
from hugging face or civid AI if you
already have a folder where you keep all
of your models you can redirect the path
of comfy UI that looks for models to
another folder and this is how you would
do it you see how in comfy UI there's
this thing called extra modore paths.
yl. example what you're going to do is
that you're going to rename this
file rename it and you're just going to
get rid of the example part so get rid
of example press enter and then it says
use. yo now that you've renamed it
you're going to need to open this up in
a text editor now I don't know what your
default text editor is but for many of
us we could really just use um text edit
that comes with Mac if not text edit
maybe notes or whatever but you know and
there's all types of text editors you
can use brackets Visual Studio code you
can see I even have X code by default
but we're just going to use text edit
for now
and what happens is if you already have
automatic 1111 you can actually go you
can actually set the path to wherever
you're keeping your models in that
folder so let me show you what I mean by
that as you can see where I keep my
automatic 1111 is in my home directory
and then you would go to stable
diffusion D webui then you would scroll
down and I would go to models and then
boom I would go to stable diffusion and
here are all my models right but you're
not going to actually give it that whole
path what you're going to do you're just
going to come to where you have stable
diffusion D webui and you're going to
right click on this and then you're
going to hold down option and then
you're going to see copy stable Das
diffusion webui as path name and you're
going to click that okay then you're
going to come over here where it says
base path you're going to remove this
and you're going to copy the path that's
there great then you're going to save
file save boom now the extra models path
is saved now that you've changed the
path you will have to restart comfy UI
so again you're just going to come over
here you're going to ex out this bad boy
pretty simple stuff then you're going to
open up another terminal you're going to
CD your way into that folder but you can
literally just drag this folder into
there and then boom now you've cded your
way in and you're going to do Python 3
main.py and it's going to start it up
again like normal you're going to head
over over to that
link open
link and now when you when you look it
up boom you see all the models that I
already had in my automatic 1111 can now
be used here so let's try out dream
shaper right this is where we had the
dream shaper remember we didn't have the
dream shaper before so now that we have
the dream shaper now the only thing that
we don't have is this upscale model but
we could just delete those for a little
bit right now let's delete those and
we're just going to put the image here
and let's see what dream Shapers gives
us and the beautiful thing is that while
you're waiting on this if you look in
the terminal you can actually see the
percentage you can also see it here
about what's loading up um you know the
green bar is also showing you what's
loading up but if you want to see a more
specific version it says 65% 70% in the
terminal itself so that's always cool to
know to see how long it's taking for you
and then it's going to go into the
decoder and again you know it's loading
one new model and everything and in a
few seconds we should have it
ooh not
bad definitely leave a like definitely
subscribe your boy Chelli just made his
wife like this is not bad this is not
bad now of course you could always take
this to another level right add some
upscaler add some different you know um
features that you would want with this
but here we have it here we have it this
is what kyui is able to create and it's
in our control and that's the Beauty and
it wasn't that hard now was it now again
if you have a M2 or an M3 MacBook the
generation will be a little faster
especially if you have like a M3 Max and
things like that the generation will be
faster but um this is comfi in a
nutshell there are definitely some comfi
resources that you can use you see
comfyi examples it shows you like you
know image to image what that would look
like and how you would go through it
hyper networks all that types of stuff
it's pretty impressive stuff um there's
also the GitHub the GitHub also shows
you a lot of what you can do here and
also redddit as you can see here a few
examples of my comfyi look at them look
at this and you know this you know
Reddit is always a budding Community for
anyone who's trying to learn anything
right we already know this so you have
Reddit there's definitely a lot of
things that you can look at um but yeah
you know there's a lot going on here you
look a beginner's guide to come for UI I
personally like this website a lot
stable diffusion dart.com um a link will
be in the description really good
website it kind of gets you up to date
and kind of gets the ball rolling for a
lot of AI image generation oh and you
know what here's also also the settings
shows you a little a lot of different
settings I personally you don't really
need to mess around too much with these
settings but there's definitely a lot of
things in the settings that you could
look over like um it does show you
different colors that you could use so
if you use solariz and things like that
or Nord you know there's definitely
different colors me personally I just
use dark by default and there's
definitely other things here you can
always play around with this stuff but
for the most part um not not much that's
really going to change anything and um
there's definitely other stuff here but
that is the bulk of com fui for Mac I
genuinely appreciate you guys for
watching please stay tuned for more
videos about com fui sdxl stable
diffusion AI in general if you really
appreciate it remember to like And
subscribe I will appreciate that in
return and please have a great day chuku
buam Chelli out
浏览更多相关视频
日本一わかりやすいStableDiffusion WebUI AUTOMATIC1111(ローカル版)のインストール方法と基本的な使い方
Instant IDでLoRAが不要になる?【Stable Diffusion WebUIでInstant IDを使い同じ顔の人物を生成する方法】
【AI動画の作り方】アニメのai実写化動画はどう作る?SNSでバズるAI動画を作って、マネタイズする方法とは
【Web制作AI活用】プログラミング無しで綺麗なウェブサイトが出来る革新的な組み合わせ【v0 × Rweb】
Lip Sync for your AI MOVIES!
AIで機能付きWEBサイトを生成するAI『Create』が凄い!
5.0 / 5 (0 votes)